Endotracheal tube placement confirmation: 100% sensitivity and specificity with sustained four-phase capnographic waveforms in a cadaveric experimental model. Silvestri S., Brown, J. Et Al. Resuscitation 2017
James Brown, MD. trained ORMC, Orlando Regional Medical Center, and currently is working at Health Central Hospital 10 miles west of downtown Orlando. He has published the second study investigating cadavers as models for tube confirmation by waveform capnography.
Waveform capnography is one of the best tube confirmation devices we have available but when is waveform capnography unreliable? The common caveats to interpreting waveform capnography include states of low-flow. Either CO2 is not being delivered to the alveoli or the CO2 can’t escape from the alveoli to the detector device. The most commonly cited examples are cardiac arrest, massive pulmonary embolism, severe asthma exacerbations or foreign bodies/materials. These all intuitively would lower the amount of CO2 ultimately detected by your capnography device, but would it change the observed waveform? This study aimed to replicate a state of extremely low flow, or should we say zero-flow state, by using cadavers and examining their waveforms.
Methods:
Two cadavers were intubated in multiple tube positions and continuous waveform capnography was measured. Tubes were placed by video laryngoscopy into the trachea, hypopharynx or the esophagus and a total of 195 ventilations were given and measured. A positive waveform had the characteristic 4-phase alveolar waveform and there was no minimum numerical value deemed necessary to be positive.
A total of 192 tracheal ventilations, 42 esophageal intubations and 6 hypopharyngeal ventilations were given over the course of 6 hours for cadaver one and 2 hours for cadaver two. At various time points over several hours 2-10 measurements were made and three EM doctors agreed whether or not a waveform was present. The tube was retracted at several time points to where the balloon was inflated superior to the glottis but making a seal with the glottis while the tip of the tube and Murphy’s eyelet remained within the trachea. These were also considered endotracheal measurements and thought to represent the minimum depth for ventilation.
The alveolar waveform. Phase 1: inspiratory plateau. Phase 2: expiratory upstroke. Phase 3: expiratory plateau. (*): measured ETCO2. Phase 4: Expiratory downstroke.
Results:
All endotracheal intubations delivered the characteristic 4-phase waveform confirming intubation. This included the measurements where only the tip of the tube was in the trachea. Furthermore, no intubation in the esophagus or hypopharynx demonstrated the correct waveform. While quantitative measurements decreased from initial time of intubation, 56mmHg to 2mmHg, even at low absolute values, the waveform was still present. From this the authors demonstrated a 100% sensitivity and 100% specificity for waveform confirming endotracheal intubation even in a zero-flow state.
Results:
Sensitivity 100%
Specificity 100%
Two main questions arise from this study:
Are cadavers adequate models for cardiac arrest or have they lost relevance by being frozen and unfrozen over time. Could other physiologic changes affect the results and interfere with the above findings. Secondly, at what point do we start to see an alveolar waveform while crossing the threshold from hypopharyngeal intubations to endotracheal intubations. Is the seal the necessary aspect of an intubation to produce the correct waveform? To address these and more, we have Dr. James Brown, corresponding author for the article.
Interview with author James Brown, MD:
How did you come to do this project?
This was the brainchild of my program director, Dr. Sal Silvestri. He had been looking into this topic and initially published an article in 2005 in Annals of Emergency Medicine investigating the utility of waveform capnography for confirmation of tube placement, specifically in the pre-hospital setting. He had this project in mind for about 8 years since then, before I was even a resident, and in 2010 and in 2015 the AHA recommendations specifically regarding endotracheal tube placement, essentially cautioned against its use in the setting of low pulmonary blood flow state, such as cardiac arrest.
Right, so one of the concerns is that waveform capnography has false negatives in low flow states, not necessarily as much concern with false positives – your project was to demonstrate the ultimate low flow state by using cadavers?
That’s correct, when we initially set out to do this study, we wanted to attempt it on recently deceased individuals, ie in the minutes or hours after a code was called. But for obvious reasons that presented some difficulties as far as obtaining consent, the ethics of it, and IRB approval.
Yeah I could imagine some concerns with that, ‘let’s go ventilate this person right after they died’ could definitely be problematic.
Exactly, exactly and that’s ultimately why the gestation period of this project was so long. It took years. Then there was a study out of Australia by Cliff Reid that showed freshly thawed cadavers were able to produced sustained alveolar capnographic waveforms.
Was there concern that thawed cadavers wouldn’t produce accurate results compared to fresh cadavers?
It was hard to even procure the cadavers and it wasn’t until recently that we even had the opportunity to do it. Essentially, we had to use a cadaver that was already being used for an IO placement lab. So, we were actually in there manipulating the head and neck of the cadaver while the labs were going on. Those IO labs at our institution, only come once or twice a year so we had to wait. We had already been approved by the company that procures the cadavers and the company that does the IO to do the study.
So, one of the things I saw on the graphic representation is that your CO2 numerical value starts really high and then tapers off, then is high again and tapers off, how do you explain that? Why wasn’t it one slow slope coming down?
Truth is we really don’t know. Our best explanation is that the CO2 measurements were from a fermentation process that was happening within the cadavers themselves. There are couple references that might explain those findings because they do start fairly high, drop off and start back up again. That’s something I think an advanced physiologist could answer and something we address in our limitations. Most likely the increases we’re seeing are probably part of the fermentation process. However, the exact physiologic explanation for this was never the goal of our study.
The aspect of this that really holds up though, is that there is really only one physiologic mechanism that can produce an alveolar waveform like that and its alveolar air mixing with dead space air within the small airways. Even though we’re acknowledging that those capnographic waveforms don’t arise from true gas exchange, the cadavers still produced these sustained waveforms for many hours.
How exactly does your study compare to the Cliff Reid study? Was this a project in parallel or were there aspects you were trying to grow on?
The Cliff Reid study is a demonstration that cadavers can produce capnographic waveforms; the next step was a controlled experiment like we attempted. We’re essentially replicating that study but I think the way our study compares is through expansion and experimental application of the same concept.
So, do you feel like your study is the definitive study to say that a proper capnographic waveform, even in arrest, always confirms tube placement with the same sensitivity and specificity as you found in cadavers?
I do. That’s essentially our conclusion. Of course, I believe our findings should be replicated and we encourage that. Mainly, the false negatives were the biggest dilemma of the whole concept. The point that we’re trying to make is to highlight the complacency involved with those false negatives. We reference the NAP4 review which showed that some unintentional esophageal intubations were ignored in the setting of cardiac arrest because of physicians essentially blaming cardiac arrest as causing the lack of waveform. Addressing that is the spirit of this whole paper.
Some of the earlier studies with the high false negative rates don’t have the same emphasis on the waveform even though the study was on capnography. Sometimes they were still using the numerical value associated with capnography readings.
Some of the misconceptions from the earlier papers is that we’re talking about canography and not capnometry and those are often confused. Many of those studies were relying on a number and not a visual waveform and I think that’s where some of the misconceptions in the AHA guidelines come from.
What are other pitfalls with waveform? Is there any way you can see the alveolar waveform and not have a confirmed tube?
I could see some misinterpreting a waveform from preserved CO2 in the esophagus, but it shouldn’t be an actual alveolar waveform. In all the esophageal intubations we had, they produced essentially a flat line with a couple blips and that could potentially produce some confusion.
I’d also like to add that our study suggests that capnography is the gold standard for verification of endotracheal placement, but should not replace basic fundamentals of post-intubation care to prevent right main-stem intubations, etc. i.e. clinicians should still listen for bilateral breath sounds and obtain chest radiographs.
Anecdotally I’ve seen a nasotracheal tube that was placed in the hypopharynx produce a 4-phase waveform. The person had a pulse but the ETCO2 measurements were about 10mmHg. I took a look with video and it was definitely not endotracheal. Is there still room for false positives with hypopharyngeal intubations?
We didn’t see any waveforms with hypopharyngeal waveforms but we didn’t have that many measurements. Our supraglottic cuff trials try to address that. We had the balloon inflated right above the glottis to make a seal but the tube and eyelet were still endotracheal. That was our method of addressing that. Our institution is working towards a similar experiment with supraglottic airway devices such as LMA and King airways in the near future. This will hopefully provide more insight into this question.
Right and that’s a tube that we wouldn’t really be that comfortable calling secure or leaving in that position even though it has the correct waveform.
Last question: at what point do you think you lose the waveform? Had the cuff been deflated, or had it not had a good seal, would some version of a waveform exist?
In our study no alveolar waveforms were produced until the cuff was inflated and occluding the glottis. While we did not obtain any endotracheal measurements with the cuff deflated, our hypopharyngeal measurements were taken with the tube bevel literally just millimeters above the glottis (picture below). There were aberrations to the waveform, but no sustained alveolar waveform, and the capnometric value remained zero.
Tube position during hypopharyngeal measurements
Thanks James!
Terren Trott (@tsquaredmd)
Sustained life-like waveform capnography after human cadaveric tracheal intubation. Cliff Reid et al. EMJ 2015, 32 (3): 232-3
Does end-tidal capnography confirm tracheal intubation in fresh-frozen cadavers? Cliff Reid et al. EMJ 2017, 34 (5): 315-317
5minuteairway 